Medical Science Educator Образование

Workflow Design, Editor Expertise and case Typology for AI-Generated OSCE Scripts

Mu, Zhanglei
Springer Nature · 2026-12-01 · DOI 10.1007/s40670-026-02935-w

Аннотация

The pilot comparability study by Josan and colleagues comparing GPT-4-generated and expert-written OSCE standardized patient scripts offers timely insights for medical education. This letter highlights three considerations. First, the “AI-only” arm was not human-free but reflected an expert-supervised iterative loop, suggesting that the key contrast lies in workflow design rather than authorship. Second, the AI+human scripts scored highest in clinical accuracy yet lowest in realism, consistent with evidence that combined human-AI workflows can underperform the stronger party alone, particularly when humans act as final arbiters rather than co-producers. Third, large language models may be better suited to knowledge-intensive, protocol-driven OSCE cases than to communication-heavy scenarios. Future studies should distinguish subject-matter from task-specific editor expertise and develop a case typology to guide appropriate AI deployment.

Читать у издателя →